Using computational models to aid drug discovery

On average, bringing medicines to market is now costing upwards of $2 billion. A company might succeed where others fail simply by virtue of a shorter development timeline and lower costs. Of course, it is often possible to increase the speed of development by simply throwing money at a problem, something which has become increasingly common due to first-in-class advantages. However, if this is done without giving due consideration to efficiency and risk, then in the long term it is not sustainable and may even lead to reputational damage if drugs produce health concerns or environmental impacts.

The use of vertebrate animals for research purposes is very tightly regulated by law, driving up the costs of the experiments, and severely limiting the types of experiments that can be undertaken and the range of conditions that can be applied. There is a very high level of variability between sites which makes experimental outcomes non-reproducible. And there is also increasing societal pressure to reduce the use of animals in research. These factors have combined to spur growing interest in the so called 3 “Rs” – reduction, refinement and replacement – in animal experimentation, across academia and industry.

In the UK alone, the chemical industry spends tens of millions of pounds each year conducting toxicology tests on fish. Standard toxicology tests, which include a fish experiment component, can take anything from a few days to one year to conduct, depending on complexity, scope and the mode of action of the drug or chemical being tested. Active pharmaceutical ingredients (APIs) are excreted and enter the environment, where they can stay in ecosystems for years. Due to the environmental impact, regulations require that before authorisa tion, new APIs must undergo an environmental risk assessment. Being able to accurately predict the risk of environmental harm far earlier in the drug development process would dramatically reduce the cost of regulatory compliance and permit better risk management and strategic decision-making in the drug industry.

In response to this need, the ‘Virtual Fish EcoToxicology Laboratory’ is to be developed by SimOmics, a spin-out established with the support of The Royal Academy of Engineering. The ‘virtual lab’ will be a transparent, evidence-based system of interlinked mathematical models, combined with extensive datasets to predict the environmental impact of APIs on everything from fish reproduction to fish behaviour, using computer simulations. This will permit a generation of evidence-based quantitative risk assessments for individual drugs in a unique web-based tool. It may also have the potential to simulate the effect of future drugs on humans before clinical trials ever take place.

Computer modelling can also help improve speed and efficiency in the pharmaceutical industry. In recent years, the value of mathematical and computational models has been widely demonstrated, particularly in the area of pharmacokinetic/pharmacodynamic (PK/PD) modelling used to understand how the body distributes and metabolises drugs.

Technology capable of providing credible and informative models to the pharmaceutical industry is a fast growing area and where those models are executed rigorously, they have transformative potential for increasing profitability. Computer modelling has the potential to help identify suitable candidate drugs faster, and carry out risk assessment and optimisation far earlier in the development cycle. The quality of these models has, in the past, been highly variable both in terms of capturing reality and in the transparency of the evidencing. Transparent evidencing is vital to enable the value of any particular model to be accurately assessed and incorporated into decisions with the appropriate weight.  

Preclinical and clinical stages of drug development are notoriously expensive, presenting remarkable opportunities for technology that can reduce the number, size and length of experiments or clinical trials. Modelling can also help identify biomarkers and end-points that are most likely to show clear evidence of therapeutic efficacy in clinical trials.

Evidenced-based mathematical and computational simulation has the potential to revolutionise the drug development and therapeutic pipeline by significantly reducing the number of unnecessary animal experiments and accelerating the translation into human clinical trials. In the future, ‘simulations’ will have a vital role to play in developing ‘personalised medicine’, using computerised clinical trials on ‘virtual patients’ to predict how each individual person may respond to a particular therapy thus reducing risk and saving significant cost.  

Paul Andrews, Ed Clark, Becky Naylor, Jim Walsh and Mark Coles, SimOmics Ltd, also contributed to this article.